Total Pregnancies in the United States, 1974-2014

Introduction

The total number of pregnancies per year is rarely investigated in demographic and fertility studies. However, this quantity is estimated herein as an intermediate first step on the way to estimating the number of unintended pregnancies per year in the United States, which is a variable of great interest in the study of reproductive responsibility.

The total number of pregnancies includes pregnancies of all outcomes, including live birth, induced abortion, miscarriage, stillbirth, and ectopic pregnancy.¹ The number of miscarriages, however, is almost surely an underestimate. Some proportion of miscarriages that occur early in pregnancy are never detected and never known to the woman carrying the pregnancy. Such pregnancies are thus not counted.

While how great of a proportion of miscarriages are thus undetected is an interesting topic for other fields, it is not particularly germane for investigations in reproductive responsibility. Therefore, this investigation leaves the number of such cases as a known unknown.

Data Source

The estimates presented below are based on the National Survey of Family Growth (NSFG) from Cycle 3 of the survey to the 2015-2019 release of the survey. Cycle 1 and Cycle 2 are not used because these cycles only surveyed women who were ever married or single and living with a child of their own in the same household. These cycles thus missed many women who experienced pregnancies, but were neither married nor cohabitating with a child of their own.

Because the NSFG is known to have a large under-reporting of pregnancies ending in induced abortion, specially adjusted weights were used. These weights are designed such that estimates of pregnancies that end in live births and estimates of pregnancies that end in induced abortion match numbers from external sources.

The same procedure for weight adjustment to compensate for abortion under-reporting, which preserved and replicated existing post-stratification, was used for all releases of the NSFG. For most releases of the NSFG, this reduced the estimate error for pregnancies counts, as would be expected for such an adjustment.

However, the public use data file for Cycle 4 of the NSFG had a post-stratification scheme for its weights that was more complicated than the post-stratification done for other releases of the NSFG, and the weight adjustment actually increased estimate error. The result is that the confidence intervals for estimates based on Cycle 4 of the NSFG – i.e., for years 1982, 1983, 1984, and 1985 – are noticeably larger than other years.

Pregnancies are counted in the year that the pregnancy ends. Thus, if a pregnancy is conceived in September 1984 and continues until June 1985, it is counted as being a pregnancy in 1985.

Results

As seen in Figure 1, the number of pregnancies per year in the United States increased substantially in the second half of the 1970s. This is consistent with this being a time of maximal increase in the number of women of reproductive age in the United States.

After this period, there was a plateau in the number of pregnancies in the first half of the 1980s. After this, there appear to be fluctuations, but the total number of pregnancies appears to stay close to 6.25 million per year throughout the rest of the analysis period.

The fluctuations in estimates of total pregnancies appear similar to the fluctuations in the number of live births. For instance, there was a local maximum in live births per year in 1990, and there is a local maximum in estimates of total pregnancies per year in Figure 1. This should not be surprising as the external counts of live births were incorporated into the estimates of total pregnancies by way of the weight adjustments.

Estimates of pregnancies occurring in girls under 15 years of age are not depicted in a subplot in Figure 2 because for most years these estimates are not different from 0 with statistical significance. A similar phenomenon occurs for women aged 40 years and over, but that subplot is included in Figure 2 because there are several years with nonzero estimates.

Estimates of number of total pregnancies by age group in Figure 2 generally mirror counts of live births by age group, but with a greater number of total pregnancies than number of live births in any given year, as expected. While it cannot be discerned, because of insufficient sample sizes of this age group in the NSFG, whether there is an increasing trend in total pregnancies among women 40 years and over, all other age groups undergo trends in total number of pregnancies similar to trends in live births.

Pregnancies and live births among teens and among women aged 20-24 had generally declining trends during the analysis period. Pregnancies and live births among women aged 25-29 years increased in the 1970s and then plateaued during the rest of the analysis. There was an increasing trend in both pregnancies and live births among women aged 30-34 years and among women aged 35-39 years from 1974 through 2014.

Footnote